Papers by Mikko Kurimo
Morfessor EM+Prune: Improved Subword Segmentation with Expectation Maximization and Pruning (2020.lrec-1)
Copied to clipboard
| Challenge: | Subword segmentation is a standard preprocessing step in many neural approaches to natural language processing. |
| Approach: | They propose to train a unigram subword model using a recursive algorithm and lexicon pruning algorithm. |
| Outcome: | The proposed method improves on the original training algorithm and improves morphological segmentation accuracy. |
When to Laugh and How Hard? A Multimodal Approach to Detecting Humor and Its Intensity (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods to generate humor using multimodal data are needed to study the role of humor in human social function. |
| Approach: | They propose a model that automatically detects humor in the Friends TV show using multimodal data and use prerecorded laughter as annotation as it marks humor. |
| Outcome: | The proposed model detects humor 78% of the time and how long the audience’s laughter reaction should last with a mean absolute error of 600 milliseconds. |
Collecting Linguistic Resources for Assessing Children’s Pronunciation of Nordic Languages (2024.lrec-main)
Copied to clipboard
Anne Marte Haug Olstad, Anna Smolander, Sofia Strömbergsson, Sari Ylinen, Minna Lehtonen, Mikko Kurimo, Yaroslav Getman, Tamás Grósz, Xinwei Cao, Torbjørn Svendsen, Giampiero Salvi
| Challenge: | Using annotated corpora of languages is difficult for children learning a foreign language . most effort is directed to the most popular languages and adult learners . |
| Approach: | They collect annotated corpora of languages spoken by children in three Nordic countries . they hope to make the data available for future research . |
| Outcome: | The collected data will be used to develop and evaluate computer assisted pronunciation assessment systems for non-native children learning a Nordic language (L2) and for L1 children with speech sound disorder (SSD). |